Update app.py

app.py

@@ -1,30 +1,79 @@
-# app.py — Space 4
-# requirements.txt: transformers, torch, gradio
-
 import gradio as gr
-from transformers import pipeline
+import joblib
+from huggingface_hub import hf_hub_download
+
+# ── Your model repositories (fix typos if needed later)
+EN_REPO  = "E-motionAssistant/Englsih_Trained_Model_LR"
+SI_REPO  = "E-motionAssistant/SInhala_Text_Emotion_Recognition_Model"
+TA_REPO  = "E-motionAssistant/Tamil_Emotion_Recognition_Model"
 
-english_clf = pipeline("text-classification", model="E-motionAssistant/Englsih_Trained_Model_LR")   # ⚠ fix full name
-sinhala_clf = pipeline("text-classification", model="E-motionAssistant/SInhala_Text_Emotion_Recognition_Model")  # ⚠ fix full name
-tamil_clf   = pipeline("text-classification", model="E-motionAssistant/Tamil_Emotion_Recognition_Model")
+# Load function (cached by Gradio)
+@gr.cache_resource
+def load_model(repo_id):
+    tfidf = joblib.load(hf_hub_download(repo_id, "tfidf_vectorizer.joblib"))
+    clf   = joblib.load(hf_hub_download(repo_id, "logreg_model.joblib"))
+    le    = joblib.load(hf_hub_download(repo_id, "label_encoder.joblib"))
+    return tfidf, clf, le
+
+# Load all three (will download only once)
+en_vec, en_clf, en_le = load_model(EN_REPO)
+si_vec, si_clf, si_le = load_model(SI_REPO)
+ta_vec, ta_clf, ta_le = load_model(TA_REPO)
 
 def classify_english(text):
-    return english_clf(text)[0]
+    if not text.strip():
+        return {"error": "කරුණාකර ටෙක්ස්ට් එකක් ඇතුලත් කරන්න"}
+    X = en_vec.transform([text])
+    pred = en_clf.predict(X)[0]
+    emotion = en_le.inverse_transform([pred])[0]
+    return {"emotion": emotion, "confidence": float(en_clf.predict_proba(X).max())}
 
 def classify_sinhala(text):
-    return sinhala_clf(text)[0]
+    if not text.strip():
+        return {"error": "කරුණාකර ටෙක්ස්ට් එකක් ඇතුලත් කරන්න"}
+    X = si_vec.transform([text])
+    pred = si_clf.predict(X)[0]
+    emotion = si_le.inverse_transform([pred])[0]
+    return {"emotion": emotion, "confidence": float(si_clf.predict_proba(X).max())}
 
 def classify_tamil(text):
-    return tamil_clf(text)[0]
+    if not text.strip():
+        return {"error": "கருணையுடன் உரையை உள்ளிடவும்"}
+    X = ta_vec.transform([text])
+    pred = ta_clf.predict(X)[0]
+    emotion = ta_le.inverse_transform([pred])[0]
+    return {"emotion": emotion, "confidence": float(ta_clf.predict_proba(X).max())}
+
+# ── UI with tabs
+with gr.Blocks(title="Multilingual Emotion Recognition") as demo:
+    gr.Markdown("""
+    # Multilingual Emotion Detector  
+    ඉංග්‍රීසි / සිංහල / தமிழ் භාෂාවලින් හැඟීම් හඳුනාගැනීම  
+    (TF-IDF + Logistic Regression models)
+    """)
 
-with gr.Blocks() as demo:
-    gr.TabbedInterface(
+    with gr.TabbedInterface(
         [
-            gr.Interface(fn=classify_english, inputs=gr.Textbox(label="English Text"), outputs=gr.JSON(), title="English"),
-            gr.Interface(fn=classify_sinhala, inputs=gr.Textbox(label="Sinhala Text"),  outputs=gr.JSON(), title="Sinhala"),
-            gr.Interface(fn=classify_tamil,   inputs=gr.Textbox(label="Tamil Text"),    outputs=gr.JSON(), title="Tamil"),
+            gr.Interface(
+                fn=classify_english,
+                inputs=gr.Textbox(label="English Text", placeholder="I'm feeling really happy today!"),
+                outputs=gr.JSON(label="Result"),
+                examples=["I hate this weather", "This is the best day ever", "Why is everything so boring?"]
+            ),
+            gr.Interface(
+                fn=classify_sinhala,
+                inputs=gr.Textbox(label="සිංහල ටෙක්ස්ට්", placeholder="මම අද ගොඩක් සතුටින් ඉන්නවා"),
+                outputs=gr.JSON(label="ප්‍රතිඵලය"),
+                examples=["මට බයයි", "ජීවිතේ ලස්සනයි", "කොහොමද මේක?"]
+            ),
+            gr.Interface(
+                fn=classify_tamil,
+                inputs=gr.Textbox(label="தமிழ் உரை", placeholder="இன்று மிகவும் சந்தோஷமாக இருக்கிறேன்"),
+                outputs=gr.JSON(label="முடிவு"),
+                examples=["எனக்கு கோபமாக இருக்கு", "இது சூப்பர்!", "எல்லாம் சோர்வாக உள்ளது"]
+            ),
         ],
-        ["English Emotion", "Sinhala Emotion", "Tamil Emotion"]
+        tab_names=["English", "සිංහල", "தமிழ்"]
     )
 
 demo.launch()